Sei sulla pagina 1di 40

Clock Domain Crossing &

Asynchronous FIFO
Basics
Introduction
• Why we need Asynchronous FIFO?
– To safely pass data (multi-bit) from one clock domain to another clock
domain.
– Basically it’s a Synchronization mechanism.
• What will happen if we don’t use any kind of Synchronization
Techniques
– Metastability
• Output of the receiving flip-flop goes to metastable state.
• Which causes propagation of invalid data throughout the design.
• Synchronization Failure !!
First – We will try to pass single bit
signal !!

• Synchronization failure => occurs when a signal generated in one


clock domain is sampled too close to the rising edge of a clock signal
from another clock domain.
Solution - Synchronizers
• Two flip-flop synchronizer
• 1st FF will go to metastable
state.
• 2nd FF will stabilize the data

• But, it depends on how fast


the data is changing
• If data is changing to
fast we might have to
consider adding 1
more FF.
MTBF (Mean Time Before Failure)
• Figure of merit for a FF related to metasabilty.
• MTBF – Basic Idea !!
– Inversely proportional
• Frequency of receiving clock domain 1
• The rate of change of data 𝑀𝑇𝐵𝐹 =
𝑓 𝑟𝑒𝑐𝑒𝑖𝑣𝑖𝑛𝑔 𝑐𝑙𝑜𝑐𝑘 ∗ 𝑓 𝑑𝑎𝑡𝑎
Problem Scenario
• Passing a Signal (1 bit) – Using 2 FF Synchronizers
From To Issues !!
Slower Clock Faster Clock Not a problem
The faster clock
signal will sample the slower
signal one or more times

Faster Clock Slower Clock There are some


exceptions
• Solution 1 – Signal
must be wider than the
cycle time of the slower
clock
• Solution 2 – Feedback
Passing a Signal from Faster clock to
Slower Clock
• What is the problem!!
– Signal will not be captured into the slower clock domain
Solution 1 – Open loop Solution
• To assert the signal for a period of time that exceeds the time period of the
sampling clock (1.5 x Time period of the receiving clock ).
– The assumption is that the control signal will be sampled at least once
by the receiver clock.
– Frequency of the receiving clock and pulse width of the signal must be
know.
Solution 2 – Closed loop Solution
(Feedback loop)
• To assert a signal, synchronize it into the new clock domain and then pass the
synchronized signal back through another synchronizer into the sending clock
domain as an acknowledge signal.
• Then, put some logic to check the data is properly captured.
– Safe Technique.
– But, there is considerable delay associated with synchronizing the signal in
both directions before releasing it.
Passing Multiple Control Signals
• The importance of the sequencing of the control signals.
– If the order or alignment of the control signals is significant, care must
be taken to correctly pass the signals into the new clock domain.

• Problem 1 - Two simultaneously required control signals.


• Problem 2 - Two phase-shifted sequencing control signals.
• Problem 3 - Two encoded control signals.
Data-Path Synchronization
• Passing data from one clock domain to another.
– Using synchronizers to handle the passing of data is generally
unacceptable.
– Why ??
• High probability for Multi-bit data changes to be incorrectly
sampled using synchronizers.
• Solution
– Solution 1 : Use handshake signals to pass data between clock
domains.
– Solution 2 : Asynchronous FIFO - store data using one clock domain
and to retrieve data using another clock domain.
Handshaking Data Between Clock
Domains
• The sender places data onto a data bus and then synchronizes a
"data_valid" signal to the receiving clock domain.
• When the "data_valid" signal is recognized the receiver clocks the data into
a register
• Receiver passes an "acknowledge" signal through a synchronizer to the
sender.
• When the sender recognizes the synchronized "acknowledge" signal, the
sender can change the value being driven onto the data bus.
• Third control signal, "ready” - to indicate that the receiver is indeed
"ready" to receive data.
• Disadvantage - The latency required to pass and recognize all of the
handshaking signals for each data word
Passing Data By FIFO Between
Clock Domains
• Most popular methods of passing data between clock domains.
• Asynchronous FIFO – Basic Idea !!
– Data are written to a FIFO buffer from one clock domain.
– Data are read from the same FIFO buffer from another
clock domain.
• Asynchronous FIFO Design – Deals With !!
– Generating FIFO Pointers
– Determining Full and Empty Status
Synchronous FIFO
• FIFO where writes to, and reads from the FIFO buffer are
conducted in the same clock domain.
• Implementation
– Using a single counter
Counting the number of writes to, and reads from the FIFO buffer.
• Increment - on FIFO write but no read
• Decrement - on FIFO read but no write
• Hold - no writes and reads
• Empty - FIFO counter = 0
• Full - FIFO counter = depth of the FIFO
Synchronous FIFO
• Using Write Pointer and Read Pointer
– Write Pointer
• Points to the next word to be written
• on reset - set to zero
– Which also happens to be the next FIFO word location to be
written.
• on FIFO- write operation
– The memory location that is pointed to by the write pointer is
written.
– Then, the write pointer is incremented to point to the next
location to be written
Synchronous FIFO
• Using Write Pointer and Read Pointer
– Read Pointer
• Points to the current FIFO word to be read
• on reset - set to zero
– FIFO is empty
– The read pointer is pointing to invalid data
• If the write pointer increments .
– The empty flag is cleared
– The read pointer that is still addressing the contents of the first
FIFO memory word
– Immediately , drives the that first valid word onto the FIFO
data output port
Synchronous FIFO
– Using Write Pointer and Read Pointer
Basically two counters running at same clock
• Increment (Write Pointer ) - on FIFO write but no read
• Increment (Read Pointer) - on FIFO read but no write
• Hold (Both) - no writes and reads
• Empty - Write Pointer - Read Pointer == 0
• Full - Same ??
– Technique used to distinguish between full and empty
– To add an extra bit to each pointer
Technique used to distinguish between
full and empty
• Add an extra bit to each pointer.
• When the Write pointer increments past the final FIFO address –
– The write pointer will increment the unused MSB .
– i.e. FIFO has wrapped and toggled the pointer MSB
• Same as for Read pointer
• If the MSBs of the two pointers are different –
– It means that the write pointer has wrapped one more time that the read
pointer
– Full Condition.
• If the MSBs of the two pointers are the same
– It means that both pointers have wrapped the same number of times
– Empty Condition
Asynchronous FIFO Pointers
• The increment-decrement FIFO fill counter cannot be used.
• Two different and asynchronous clocks (wclk & rclk) would be
required to control write pointer and read pointer.
• To determine full and empty status, the write and read pointers will
have to be compared.

• As ,Write pointer and Read pointer


– Uses n bit.
– Then,(n-1) is the number of address bits required to access the
entire FIFO memory buffer.
– Therefore ,The FIFO design uses n-bit pointers for a FIFO with
2 power (n-1) write-able locations (depth).
Problem with Binary FIFO pointer
• Trying to synchronize a binary count value from one clock
domain to another is problematic.
– Because, every bit of an n-bit counter can change simultaneously
– 7 -> 8 in binary numbers is 0111->1000, all bits changed.
– The synchronized and sampled write & read pointer might not reflect
the current value of the actual write & read pointer.
• Solution - Gray code counters
– Gray codes only allow one bit to change for each clock transition,
– eliminating the problem associated with trying to synchronize multiple
changing signals on the same clock edge.
Gray code counter
• Gray code – Unit distance code !!
• The problem of converting an n-bit Gray code to an (n-1)-bit Gray code.
Problem of converting an n-bit Gray
code to an (n-1)-bit Gray code.
• Observe :
– The second half of the 4-bit Gray code is a mirror image of
the first half with the MSB inverted.
– To convert a 4-bit to a 3-bit Gray code - we want the LSBs
of the second half to repeat the 4-bit LSBs of the first half.
But, This creates another problem!!
- Two bits are changing !!
- A true Gray code only changes one
bit between counts.
Solution - Dual n-bit Gray code counter
Style 1
• Create both an n-bit Gray code count and an (n-1)-bit Gray
code count.
– n-bit Gray code sequence - Write / Read pointer.
– (n-1)-bit Gray code sequence – to address memory.
• Dual n-bit Gray code counter – Style 1
Solution - Dual n-bit Gray code counter
Style 1
• Dual n-bit Gray code counter – Style 1
– Generates both an n-bit Gray code sequence and an (n-1)-bit Gray code
sequence.
– The (n-1)-bit Gray code is simply generated by doing an exclusive-or
operation on the two MSBs of the n-bit Gray code to generate the MSB
for the (n-1)-bit Gray code.
– This is combined with the (n-2) LSBs of the n-bit Gray code counter to
form the (n-1)-bit Gray code counter
• Drawback - (n-1)-bit Gray code sequence (to address memory)
to binary values.
Solution - Dual n-bit Gray code
counter Style 2
Solution - Dual n-bit Gray code
counter Style 2
• Employs two sets of registers to eliminate the need to translate
Gray pointer values to binary values.
• The second set of registers (the binary registers) can also be
used to address the FIFO memory directly without the need to
translate memory addresses into Gray codes.
Handling full & empty conditions
• Empty flag will be generated in the read-clock domain
– To insure that the empty flag is detected immediately
when the FIFO buffer is empty.
– i.e. the instant that the read pointer catches up to the
write pointer (including the pointer MSBs).
• Full flag will be generated in the write-clock domain
– To insure that the full flag is detected immediately when
the FIFO buffer is full.
– i.e. the instant that the write pointer catches up to the
read pointer (except for different pointer MSBs).
Synchronized pointer comparison
• Write pointer and read pointer must be synchronized into
opposite clock domain.
Generating empty
• The FIFO is empty when the read pointer and the synchronized write
pointer are equal.
– If the extra bits of both pointers (the MSBs of the pointers) are equal, the
pointers have wrapped the same number of times and if the rest of the read
pointer equals the synchronized write pointer, the FIFO is empty.
• The Gray code write pointer must be synchronized into the read-
clock domain through a pair of synchronizer registers.
• In order to efficiently register the rempty output, the synchronized
write pointer is actually compared against the rgraynext (the next
Gray code that will be registered into the rptr).
• Example
assign rempty_val = (rgraynext == rq2_wptr);
always @(posedge rclk or negedge rrst_n)
if (!rrst_n) rempty <= 1'b1;
else rempty <= rempty_val;
Generating full
• Full flag is generated in the write-clock domain.
– By, running a comparison between the write and read pointers.
– The read pointer be synchronized into the write clock domain
before doing pointer comparison.

• Full comparison has issues !!


– Directly, using Gray code counters with an extra bit to do the
comparison is not valid to determine the full condition.
– The problem is that a Gray code is a symmetric code except for
the MSBs
Problems associated with extracting a 3-
bit Gray code from a 4-bit Gray code
• Example – 8 bit deep FIFO
• FIFO is allowed to fill seven
bit locations(0-6)
• Then, FIFO is emptied.
• Now, both pointers = gray 7
• FIFO empty
• But, the on the next write
operation.
• Write pointer will increment
• Making -

• Which is Wrong !!
• As, FIFO is not actually full
• And 3 LSB’s didn’t change
• Which overwrite the same
location twice
Generating full – Correct Method
• Three conditions that are all necessary for the FIFO to be full
1. The write pointer and the synchronized read pointer MSB's must be not
equal
2. The write pointer and the synchronized read pointer 2nd MSB's must
not equal
3. All other write pointer and synchronized read pointer bits must be
equal.
• Similarly, In order to efficiently register the wfull output, the
synchronized read pointer is actually compared against the
wgnext
• Example
assign wfull_val = ((wgnext[ADDRSIZE] !=wq2_rptr[ADDRSIZE] ) &&
(wgnext[ADDRSIZE-1] !=wq2_rptr[ADDRSIZE-1]) &&
(wgnext[ADDRSIZE-2:0]==wq2_rptr[ADDRSIZE-2:0]));
Asynchronous FIFO Block Diagram
FIFO memory buffer
• Dual-port, synchronous memory
• Synchronous Write
• Asynchronous Read
Read pointer & empty generation
logic
• The read pointer is a dual n-bit Gray code counter.
• The n-bit pointer ( rptr ) is passed to the write clock
domain through the sync_r2w module
• The (n-1)-bit pointer ( raddr ) is used to address the
FIFO buffer.
• The FIFO empty output is registered and is asserted
on the next rising rclk edge when the next rptr value
equals the synchronized wptr value.
• This module is entirely synchronous to the rclk
Write pointer & full generation logic
• The write pointer is a dual n-bit Gray code counter.
• The n-bit pointer ( wptr ) is passed to the read clock
domain through the sync_w2r module.
• The (n-1)-bit pointer ( waddr ) is used to address the
FIFO buffer.
• The FIFO full output is registered and is asserted on
the next rising wclk edge when the next modified
wgnext value equals the synchronized and modified
wrptr2 value (except 1st MSBs and 2nd MSBs).
• This module is entirely synchronous to the wclk
Almost full and almost empty –Different
Ways!!
• When the difference between the pointers is smaller than
the programmed difference, the corresponding almost full
or almost empty bit is asserted.
• Can also be implemented using fixed difference.
• When the MSBs of the FIFO pointers are close.
• The almost full condition could be described as the
condition when (write pointer+4) catches up to the
synchronized read pointer.
Issues with FIFO design
• Full and Empty removal is pessimistic(not accurate).
– “full” and “empty” are both asserted exactly on time but
removed late.
• Setting either the full flag or empty flag might not be quite
accurate if both pointers are incrementing simultaneously.
Reference
• Clifford E. Cummings, “Simulation and Synthesis Techniques for
Asynchronous FIFO Design,” SNUG 2002 (Synopsys Users Group
Conference, San Jose, CA, 2002) User Papers, March 2002.
• Clifford E. Cummings, “Synthesis and Scripting Techniques for Designing
Multi-Asynchronous Clock Designs,” SNUG 2001 (Synopsys Users Group
Conference, San Jose, CA, 2001) User Papers, March 2001.